The Privacy Paradox: Explained using the
Construal Level Theory
Name: Maarten Herbrink
Program: MSC Business Administration
Student number: 10970045
Date: 28-01-2016
Thesis Maarten Herbrink – The Privacy Paradox: Explained using the Construal Level Theory 2
STATEMENT OF ORIGINALITY
This document is written by Student Maarten Herbrink who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Thesis Maarten Herbrink – The Privacy Paradox: Explained using the Construal Level Theory 3
TABLE OF CONTENTS
1: INTRODUCTION 5
2: THEORETICAL FRAMEWORK 8
2.1PRIVACY PARADOX 8
2.2CONSTRUAL LEVEL THEORY 13
3: OVERVIEW OF STUDIES 21
4: STUDY 1 22
4.1METHOD 22
4.1.1RESPONDENTS 22
4.1.2DESIGN STUDY 22
4.1.3STIMULI AND MEASUREMENTS 23
4.1.4PROCEDURE 24
4.1.5DATA ANALYSIS 25
4.2RESULTS 26
4.2.1PSYCHOLOGICAL DISTANCE 26
4.2.2CONSTRUAL LEVEL MEASURE 31
4.3DISCUSSION 33 5: STUDY 2 35 5.1METHOD 35 5.1.1RESPONDENTS 35 5.1.2DESIGN STUDY 35 5.1.3MEASUREMENTS 36 5.1.4PROCEDURE 38 5.1.5DATA ANALYSIS 39 5.2RESULTS 40 5.3DISCUSSION 41 6. GENERAL DISCUSSION 42 6.1IMPLICATIONS 42
6.2 LIMITATIONS AND FUTURE RESEARCH 45
REFERENCES 48
APPENDICES 56
APPENDIXA:STUDY1QUESTIONNAIREINDUTCH 56 APPENDIXB:STUDY2QUESTIONNAIREINDUTCH 59 APPENDIXC:PRIVACYCONCERNSITEMSSTUDY2 69 APPENDIXD:STUDY1DESCRIPTION&MINDSETMEASURES 71
APPENDIXE:DEMOGRAPHICSSTUDY1 72
Thesis Maarten Herbrink – The Privacy Paradox: Explained using the Construal Level Theory 4
ABSTRACT
Customer insights are key nowadays for companies in order to create a competitive advantage by offering a superior customized product or service. However, as companies are gathering more and more data, concerns about consumers’ privacy are being more heavily expressed from within society. Despite these worries, consumers’ behaviour is inconsistent with their privacy concerns, since they are still disclosing their personal data online. This dichotomy between attitudes and behaviour, is called the privacy paradox. In this paper, the Construal Level Theory is used to explain the privacy paradox. This is done using two studies. The first study proves that the benefits of disclosing personal data online are perceived as psychologically (temporally & hypothetically) closer than the costs of disclosing this data. In study 1 it was also found that the costs of disclosing personal data online are construed as more abstract, opposed to the benefits of disclosure, which are construed more concrete. The second study did not show additional evidence for the before described expected effect between consumers having a more concrete mindset when disclosing personal data online, opposed to having a more abstract one when answering privacy concern questions.
Thesis Maarten Herbrink – The Privacy Paradox: Explained using the Construal Level Theory 5
1:INTRODUCTION
Nowadays the ability to gather, analyse and respond to user information is of a growing importance. To survive in competitive markets, companies build on vast quantities of information to build bonds with existing consumers and to attract new ones (Culnan & Armstrong, 1999). Marketers have the ability to get to know their
consumers, through the personal information consumers disclose online. Companies
use this consumer information to attempt to offer personalized services that will increase value, consequently consumer loyalty and a business advantage (Weill & Vitale, 2001; Barnes, 2006; McAfee & Brynjolfsson, 2012). This information consists of consumers’ identity, interests, location, activity and time (Junglas & Watson, 2006). Analysing online disclosed information by internet users enables marketers to ultra-efficiently target the exact right audiences for certain product offerings, and meet the needs and desires of their target groups perfectly (Moon, 2000). However, as marketers leverage the ability to collect, analyse and use great amounts of consumer information, major concerns have been expressed over the potential erosion of personal privacy as a result of the disclosing of personal data online (Whiting, 2002; Williams, 2002; Harris, 1996). Following this, several surveys identified personal information security and privacy as some of the most pressing and widespread concerns of those using new information technologies (Commission, 2000; Cyber Security Industry Alliance 2005; Milne, Rohm, & Bahl, 2004; National Telecommunications and Information Administration, 2000; Smith, 2005). We can state that online privacy has become an important social issue (Solove, 2007; Solove, 2011).
Thesis Maarten Herbrink – The Privacy Paradox: Explained using the Construal Level Theory 6 showing less willingness to disclose personal information online. Their behaviour is inconsistent with their concerns (Norberg, Horne & Horne, 2007; Jorstad, 2000; Heng et al., 2011). Research showed that consumers do not undertake action to control their personal information better and they keep freely providing their data on the internet (Culnan & Armstrong, 1999; Acquisti & Grossklags, 2005; Norberg, Horne & Horne, 2007). This dichotomy between privacy concerns and privacy protecting behaviour is called the privacy paradox (Taddicken, 2014; Jorstad, 2000; Norberg et al., 2007; Spiekermann, Grossklags & Berendt, 2001).
The privacy paradox – the gap between privacy concerns and
privacy-protecting behaviour – is well-established in the literature. This dichotomy is typically contributed to limited information about collection practices, limited ability to calculate the various parameters relevant to the choice, and psychological distortions of the rational decision making process (Acquisti & Grossklags, 2004). However, in this paper the Construal Level Theory is proposed as an underlying framework for explaining the privacy paradox.
The Construal Level Theory (CLT) proposes that psychological distance changes peoples’ responses to future events events by changing the way people mentally represent those events (Trope & Liberman, 2003; Trope, Liberman & Waslak, 2006; Kim & John, 2008). It also states that the psychological distance is cognitively related to each other and similarly affect and be affected by level of construal. Psychological distance is a subjective experience that something is close or far way from the self, here and now (Trope & Liberman, 2010; Trope, Liberman & Waslak, 2007). Another important assumption from the CLT is that it contends that people use increasingly higher levels of construal to represent an object as the psychological distance from the object increases (Trope & Liberman, 2010). Drawing
Thesis Maarten Herbrink – The Privacy Paradox: Explained using the Construal Level Theory 7 upon the above, the following research question is used in this paper: “When
disclosing personal data online, do users perceive the benefits as psychologically closer than the costs – are these benefits congruent with low-level construals – and how does this explain the privacy paradox?”
If the Construal Level Theory can account for the discrepancy between attitudes about privacy and the actual disclosure behaviours, this would imply an important scientific contribution. Several studies have tried to explain the privacy paradox (Acquisti, 2002; Awad & Krisnan, 2006; Norberg, Horne & Horne, 2007). However, no comprehensive underlying psychological framework has been found to date.
From a managerial perspective this paper offers a large contribution. Companies use information about customers to improve their services and products, and to design personalized offerings (Awad & Krisnan, 2006). It is important for these companies to know how consumers act on their privacy concerns (Westin, 1997). If the CLT can be used as an underlying explanation, this offers a better understanding by companies of what drives consumers in privacy disclosure situations, which is of huge value to them.
This research paper starts with the theoretical framework. In this part the constructs for this research are explained. Hereafter, the two studies conducted are explained. Per study first the method is explained, after which the results are discussed. In the end the discussion of the results is discussed. Lastly the limitations and directions for future research are provided.
Thesis Maarten Herbrink – The Privacy Paradox: Explained using the Construal Level Theory 8
2:THEORETICALFRAMEWORK
2.1PRIVACY PARADOX
Privacy has been a sensitive subject since long before the internet. Westin (1967) defined privacy as “the right of the individual to decide what information about himself should be communicated to others and under what condition”. Schoeman (1984) proposed privacy as a state or condition of limited access to individuals. A more modern definition of online privacy is: “the ability of individuals to control the flow of information and have reasonable access to data generated during a browsing session” (Williams, 2002).
Consumers are concerned about what companies know about them, how companies obtain and use personal information, and the accuracy of the information used (Equifax-Harris, 1995; Harris-Equifax, 1992; Katz & Tassone, 1990). Concern relates to the customers’ apprehension and uneasiness over the use of their personal data, use of personal data by third-parties and hacking and identity theft (Robbin, 2001; Westin, 2003 Boyd & Ellison, 2008). Consumer apprehensions also include the increase of databases, volume of collected personal data, the possibility of privacy violations, loss of control during the process of collecting, accessing, and how the information could be utilized (Culnan, 1993; Hiller & Cohen, 2001). This development is the call to increase our understanding of the attitudes and behaviours toward privacy-affecting systems (Iachello & Hong, 2007).
Although consumers are expressing concerns about possible privacy, they do not show less willingness to disclose personal information online (Norberg, Horne & Horne, 2007; Acquisti & Grossklags, 2004; Taddicken, 2014). Acquisti and Grossklags (2004) found that the vast majority of subjects’ expressed privacy concerns and still traded-off privacy for other advantages (rewards, convenience, etc.).
Thesis Maarten Herbrink – The Privacy Paradox: Explained using the Construal Level Theory 9 Acquisti and Grossklags (2005) showed that consumers often lack enough information to make privacy-sensitive decisions, and even with sufficient information, are likely to trade off long-term privacy for short-term benefits. Taddicken (2014) found that privacy concerns hardly impacts self-disclosure on social networking sites. This dichotomy between privacy concerns and actual privacy protecting behaviour is called the privacy paradox (Awad & Krisnan, 2006).
In the literature, several explanations are given for the dichotomy between privacy attitudes and behaviour. In the next section the focus is on the explanation that users could be rational agents, or non-rational agents in online disclosure situations.
Users as rational agents
One of the explanations for the privacy paradox, is that the choice users make is one between a better/customized product or service offering based on online data on the one hand and the privacy infringement that such disclosure causes on the other hand (Norberg, Horne & Horne, 2007). This means that economists would utilize an exchange framework in which the information the consumer receives from the website is of greater value than the information provided by the consumer to the site. However, this explanation does not explain why people give their phone numbers in simple transactions like, for example, in sports stores. O’Harrow (2005, p.7) noted: ‘consumers are often willing, even eagerly, part with intimate details of their lives’. There seems to be an asymmetric exchange; consumers receive limited value for providing information to a firm (Han & Maclaurin, 2002). This asymmetric exchange, whereby the consumer receives limited value for providing information to a firm, has been noted (Phelps, Nowak & Ferrell, 2000; Han & Maclaurin 2002).
Thesis Maarten Herbrink – The Privacy Paradox: Explained using the Construal Level Theory 10 Table 2.1
Influencers non-rational agent (Acquisti & Grossklags, 2004)
Acquisti (2002) formalizes the abstract economic trade-offs faced by a rational agent who were to decide between information release and information protection. For the consumer it offers benefits such as targeted offers and discounts in exchange for personal information. However, the cost of doing so is the risk of future revelation of personal information (Acquisti, 2002).
Users as Non-Rational Agents
Possible explanations explained above assume that consumers are rational agents when choosing to disclose information. We have to question however if this is really the case. When we go from abstract representations to actual observations, we note that human beings will be faced with an intricate web of trade-offs dominated by subjective evaluations and uncertainties when attempting to solve for the best privacy decision (Acquisti & Grossklags, 2004). They noted that there are several factors that could be at play when a consumer is faced with an actual disclosure situation opposed to a fictional survey. Three of those (table 2.1) are explained below.
Influencers individual Description
Limited information. The amount of information individual has access to. Is user aware of information security risks and what is knowledge of existence protective technology.
Bounded rationality Is individual able to quantify costs and benefits of revealing or hiding information?
Psychological distortions There are several factors that lead to deviations from behaviour we expect to see from rational agents.
Thesis Maarten Herbrink – The Privacy Paradox: Explained using the Construal Level Theory 11 Limited information has to with the amount of information an individual has access to: the awareness of the individual of information security risk, and what does the the knowledge of the individual about protective technologies (Acquisti & Grossklags, 2004). People can never get all the information. As a result of this people usually act on asymmetric information. This becomes a problem when taking action without full knowledge about risks to personal privacy (Varian, 1996). Varian (1996) discusses that incomplete information becomes a problem for individuals when they have to commit an action without full assessment of the associated privacy-risks. The individual may not be aware of the risks that occur by not protecting personal information or the ways to protect one self. Individuals also may assume that institutions are providing a secure platform for their actions.
Bounded rationality refers to both the ability of individual to quantify costs and benefits of revealing or hiding information – but also to the inability to process all uncertain information related to information security costs and benefits (Acquisti & Grossklags, 2004; Simon, 1972; Acquisti & Grossklags, 2005). Simon (1991) states that bounded rationality are the limits upon the ability of human beings to adapt optimally, or even satisfactorily, to complex environments. The information loss’ individuals experience when disclosing personal information online, persists for an unpredictable span of time. During this time the individual is in a position of information asymmetry (Acquisti & Grossklags, 2004; Acquisti & Grossklags, 2005). The negative possible future misuse of somebody’s personal information is of unpredictable scope and probability – and the individual is likely in a condition of bounded rationality. The potential consequences of disclosing data are hard to quantify or calculate (Noam, 1996). Thus, the decision where users decide not to protect themselves paradoxically may be considered as a rational way to react to
Thesis Maarten Herbrink – The Privacy Paradox: Explained using the Construal Level Theory 12 uncertainties – also called rational ignorance (Lemley, 2000).
Lastly, there are psychological distortions. Acquisti and Grossklags (2004) note various factors that can lead to predictable deviations from behaviour as expected from rational agents. McIntosh (1969) discussed the self-control problem: persons might impose constraints on their future behaviour even if these constraints limit them in achieving maximum utility. Spiekermann et al (2002) provided evidence for missing self-control. Also, individuals have the tendency to under-discount long-term risks and losses when acting in privacy sensitive situations (Jupiter Research, 2002). In experimental psychology and economics, the tendency to overvalue the immediate reward compared to the delayed one is also know as temporal discounting (Green, Myerson, Lichtman, Rosen, & Fry, 1996; Rachlin & Raineri, 1992). This touches upon the Construal Level Theory’s implications – which are explained in the next paragraph. According to Acquisti and Grossklags (2004) these factors may cause the dichotomy between abstractly stated attitudes and actual behaviour if above mentioned factors influence the decision process of the individual, and if their perception during an experiment is different from actual disclosure behaviour.
Thesis Maarten Herbrink – The Privacy Paradox: Explained using the Construal Level Theory 13
2.2CONSTRUAL LEVEL THEORY
One of the main goals of consumer psychology is understanding how individuals evaluate objects and events (Trope, Liberman & Waslak, 2007). Dual processing theories have emerged in the social psychology as some of the most influential models in understanding and explaining the shift in people’s evaluations of objects and social situations (Dhar & Kim, 2006). More recently however, the Construal Level Theory (CLT) – with a purely cognitive orientation – has risen to the forefront, with a focus on the level of construals that receive greater attention or weight in evaluation (Dhar & Kim, 2006). The CLT – a framework that links distance and level of abstraction – states that psychological distance is an important determinant of whether primary or secondary characteristics are used as the basis of evaluation. CLT states that psychological distance changes peoples’ responses to future events by changing the way people mentally represent those events (Trope & Liberman, 2003). The greater the psychological distance from an event, the more distant it appears and the more abstractly we would expect it to be represented (Trope, Liberman & Waslak, 2007; Freitas, Salovey, & Liberman, 2001; Trope & Liberman, 2003; Vallacher & Wegner, 1987). Representations of stimuli that are psychologically near are low level and concrete while stimuli that are psychologically distant are high level and abstract (Dhar & Kim, 2006). When people move from a concrete representation of an object to an abstract one, they omit features that by the very act of abstraction are deemed incidental. Examples are: moving from representing an object as an “cell phone” to representing it as a “communication device”, people omit information about size; moving from representing an activity as “playing football” to stating it as “having fun”, people omit the ball (Trope & Liberman, 2010). Concrete representations are represented by multiple abstractions. A cell phone could be
Thesis Maarten Herbrink – The Privacy Paradox: Explained using the Construal Level Theory 14 Table 2.2
Distinguishing high- and low-level construals (Trope & Liberman, 2003).
construed as a “small object” and “playing football” could be construed as “exercising” (Trope & Liberman, 2010). Because abstract representations impose one of many alternative interpretations – and irrelevant details are omitted or irrelevant – these representations have the tendency to be simpler, less ambiguous, more coherent, more schematic and more prototypical than concrete representations (Fiske & Taylor, 1991; Smith, 1998; Semin & Fiedler, 1988). As psychological distance increases, construals become more abstract, and as level of abstraction increases, so too does the psychological distances people envisage. Construal levels thus expand and contract one’s mental horizon. (Trope & Liberman, 2010)
CLT contains low-level construals and high-level construals. Low-level construals belong to near future situations and behaviours (Liberman & Trope, 1998; Liberman, Sagristano & Trope, 2002). These construals are relatively unstructured, rich in detail, contextualized, complex representations that include subordinate and incidental features of events (Trope, Liberman & Wakslak, 2007). Low-level construals explain the how of an action’s purpose (Trope & Liberman, 2003). High-level construals are High-Level Construal Low-Level Construal
Abstract Concrete
Simple Complex
Structured, coherent Unstructured, incoherent
Decontextualized Contextualized
Primary, core Secondary, surface
Superordinate Subordinate
Thesis Maarten Herbrink – The Privacy Paradox: Explained using the Construal Level Theory 15 used to represent distant future actions and consist more abstract dispositions and principles to explain distant future behaviours (Liberman & Trope, 1998; Liberman, Sagristano & Trope, 2002). These construals are schematic, incidental, decontextualized, simple representations that extract the gist from the available information (Liberman & Trope, 1998). High-level explain the purpose of why one does something (Vallacher & Wegner, 1989).
The dichotomy between privacy concerns and actual disclosure will be explained in this paper on the basis of the Construal Level Theory, because psychological distance shapes how people perceive the benefits and costs of disclosing their personal information. It is expected that people perceive the benefits of disclosing personal data online as psychologically closer, whereas the costs of disclosing personal data online are perceived as psychologically more distant. Looking at the benefits of disclosing personal data online, people will expect to receive these immediately. For example, when accepting cookies for a better usage experience of a website, people will expect an immediate improvement of a better website experience. The costs however, maybe indicated as “bad for my privacy” would be expected to be more psychologically distant, since it is uncertain if, how and when these negative consequences would occur. It is expected that the psychological distance is a reflection of the actual psychological distance of the benefits and costs in these online disclosure situations.
It is thus expected that the benefits of disclosing information are seen as psychologically closer, and the costs of disclosing this information as more psychologically distant. This is expected on the basis of two psychological distances: temporal and hypothetical distance. These dimensions are introduced in the next part.
Thesis Maarten Herbrink – The Privacy Paradox: Explained using the Construal Level Theory 16
Temporal and hypothetical distance
Psychological distance in CLT refers to the distance of an object from the direct experience, which consists out of different interrelated dimensions: temporal, social, spatial, and hypothetical distance (Liberman, Trope, McCrea, Sherman, 2007). These dimensions influence the mental construal. Spatial distance has to do with distant related sellers; social distance refers to the social distance perceived from sales people or customer service representatives (Trope, Liberman & Waslak, 2007). In the light of the privacy paradox, the temporal and hypothetical distance are most relevant. These two dimensions are used in this paper.
Temporal distance refers to distance in time from an event (Liberman et al., 2007). Temporal distance is seen as one of the most important determinants of the psychological distance (Trope & Liberman, 2003; Trope et al., 2007). CLT proposes that the greater the temporal distance from a future event, the more likely the event will be represented abstractly in terms of a number general features that carry the perceived essence of the events rather than in terms of concrete and more incidental details of the event (Trope & Liberman, 2003). As explained, these distant future events are generally represented in a more abstract, high-level way, whereas near future events are represented in a more concrete, low-level way (Trope & Liberman, 2003; Trope et al., 2007). Linking this to dichotomy of the concerns of people and the actual disclosure behaviour, the costs of disclosing personal data online will be perceived as more temporally distant, since the costs are expected to take place temporally further away. This opposed to the benefits, that are expected to be perceived as temporally close. For example, when individuals are asked to register on a smartphone app with their Facebook-account, benefits named might be: “direct
Thesis Maarten Herbrink – The Privacy Paradox: Explained using the Construal Level Theory 17 data are likely to happen direct on the temporal distance scale. A cost that might be named is “my Facebook-data can be misused by the company”. This is more likely to be evaluated temporally further away than the concrete benefits.
Hypothetical distance refers to the likelihood of an event occurring to people. The more likely an event is to happen, the less psychologically distant this event is perceived, and the more concrete and low-level it is perceived (Waslak & Trope, 2008; Waslak & Trope, 2009). Linking this to the dichotomy in attitudes and disclosure behaviour, it is assumed that the benefits of disclosing personal data online would be perceived as more likely to happen, which thus would mean a closer hypothetical distance. The costs however, would be perceived as hypothetically further away. For example, a benefit of accepting cookies on a website could be: “a
better usage experience”. This would be very likely to also actually happen. The costs
of accepting the cookies, maybe named “privacy infringement” is less likely to evaluate on whether it will happen.
These perceptions of psychological distance are expected to be a perception of the actual psychological distance of the benefits and costs in these kind of online disclosure situations. On the basis of previous research and current assumptions, hypothesis 1 was drawn. It is expected that people perceive the benefits of disclosing personal data online as psychologically closer (temporally and hypothetically) than the cost of doing so.
H1: In typical disclosure situations, the benefits of disclosing personal data online
are perceived as psychologically closer (temporally & hypothetically), whereas the cost of disclosing personal data online are perceived as psychologically more distant.
Thesis Maarten Herbrink – The Privacy Paradox: Explained using the Construal Level Theory 18
Construals
As explained in the previous section, the CLT contains low-level construals and high-level construals. Distant future events are construed more abstractly (high-high-level), whereas near future events are construed more concrete (low-level). Any situation people are in likely contains multiple goal-relevant cues, each of which can be construed in a myriad of ways (Brown, 1958). The most fundamental task of self-regulation is balancing abstract, long-term aims with immediate, concrete experiences, as when a smoker eschews a cigarette in return for better health (Trope & Fishbach, 2000; Rachlin, 2000). Studies (Bar-Anan, Liberman & Trope, 2006; Greenwald, McGhee, & Schwartz, 1998) showed that the association between psychological distance and construal level can be activated automatically without conscious deliberation. Research (Waslak & Trope, 2009) found that the more likely (unlikely) an event is to occur, the less (more) distant this event is construed. The mindset people have – either concrete or abstract – is thus dependent on the psychological distance perceived. Thus, when people perceive a large psychological distance, this is congruent with an abstract mindset (high-level) whereas a nearby psychological distance is congruent with a concrete mindset (low-level), since low-level construals belong to near future situations and behaviours (Liberman & Trope, 1998; Liberman, Sagristano & Trope, 2002). This congruence between psychological distance and construals affects behaviour (Trope & Liberman, 2010).
It is expected that in natural situations, in this case an online disclosure situation, the mindset of people is concrete, because people then hold low-level goals – for example a low-level goal of using a Facebook account to access a website would be: ‘to gain quick access to the website’. These low-level goals cause people to follow their proximal benefits of disclosure – such as the quick access to a website – instead
Thesis Maarten Herbrink – The Privacy Paradox: Explained using the Construal Level Theory 19 of following their distant (high-level) costs of disclosure. When thinking about privacy concerns however, which is congruent with high-level construals, people would perceive the example of using their Facebook account to access the website as: ‘bad for privacy’. In this situation it is expected that the distant costs of disclosure are followed. In hypothesis 1, it is hypothesized that people experience the benefits of disclosing personal data online as psychologically (temporally & hypothetically) closer and the costs of disclosure as more psychologically distant. As explained there is a congruence between this psychological distance and the mindset people hold – which influences behaviour. It is expected that people in online disclosure situations follow their proximal benefits of disclosing their data because people follow their low-level goals. This ‘natural disclosure situation’ can be compared to a situation in which people fill out questions about their online privacy. In this situation people are expected to follow their distant costs – congruent with an abstract mindset. This is enforced by the fact that privacy itself is an abstract concept. On the basis of these expectations, hypothesis 2 was formed:
H2: When making decisions in typical online disclosure situations, people hold a
concrete mindset, as opposed to an abstract mindset when thinking about privacy concerns
Thesis Maarten Herbrink – The Privacy Paradox: Explained using the Construal Level Theory 20
Summary
If these hypotheses can be confirmed, this will serve as evidence that the dichotomy in privacy attitudes and disclosure behaviour – the privacy paradox – can be explained on the basis of the Construal Level Theory. When in actual disclosure situations, it is expected that people perceive the benefits of disclosing their data as psychologically closer, which is shown through more concrete, low-level construals. Benefits are expected to be described in more detail than the costs. The costs are described as more abstract, and are construed on a high-level. The psychological distance is expected to be further away.
It is expected that when people are in real life disclosure situations, their mindset is congruent with the benefits, because in real-life situations, people focus on the proximal benefits of disclosing their data online, whereas their mindset is concrete because people hold low-level goals. However, when filling in surveys about their privacy concerns, the mindset of people is congruent with their distanat costs of disclosing personal data – which is why people report such high concerns about their privacy. The more concrete information is, the easier it is to imagine and associate it with affect (Newell, Mitchell, & Hayes, in press; Slovic, Monahan & MacGregor, 2000). Since the benefits are more concrete (low-level), people in real life follow the benefits of disclosure and this thus explains why there is a gap between attitudes and actual disclosure behaviour. The concerns are only shown in surveys which are experienced on an abstract level, with a high psychological distance, whereas the actual behaviour takes place on a concrete level and are congruent with a nearby psychological distance.
Thesis Maarten Herbrink – The Privacy Paradox: Explained using the Construal Level Theory 21
3:OVERVIEWOFSTUDIES
Two studies were conducted. The first study checked to see whether people perceive the benefits of disclosing personal data online as psychologically (temporally & hypothetically) closer than the costs of doing so (H1). This was tested by exposing respondents to disclosure situations and by asking them to name costs and benefits of potential disclosure, which they afterwards measured on temporal and hypothetical distance. In addition to this, an analysis was run to see whether benefits were construed more concrete than the costs, that are expected to be more abstractly construed. This also serves to test H2.
The second study tested whether people hold a concrete (low-level) mindset when making decisions about the disclosure of data, and a more abstract (high-level) one when filling in questions about their privacy concerns (H2). This was measured through a survey, in which the respondents were divided into a privacy concern group and a disclosure group. Afterwards, both groups’ their mindset was measured. The data was analysed for both studies in SPSS. Both surveys were conducted in Dutch, since this was the vast majority of the respondents’ native language.
Thesis Maarten Herbrink – The Privacy Paradox: Explained using the Construal Level Theory 22
4:STUDY1
4.1METHOD
4.1.1RESPONDENTS
In the study 117 respondents participated. Respondents were approached via Facebook, e-mail and Whatsapp. Out of the 117 people that started the survey, only 75 finished the survey, which equals a completion rate of 64%. These 75 are the respondents taken into account in the analysis. 43 respondents were male, 32 were female. Over 50% (N=38) of the respondents were between 18 and 24 years old. Looking at education 92% (N=69) of the participants are highly educated (HBO or university). The internet usage for personal use per day was for over 65% (N=51) of the respondents between 1-4 hours per day. All demographics can be found in appendix E.
4.1.2DESIGN STUDY
The goal of this study is to see how psychological distance (temporal and hypothetical) differs for the costs and benefits of certain online privacy disclosure situations. This was measured through a survey, where participants rated three online privacy disclosure situations on willingness to disclose, and afterwards self-generated costs and benefits of this possible disclosure of data, per situation. Lastly, they were asked to rate these self-generated costs and benefits on psychological (temporal & hypothetical) distance. Also, they rated how (dis)advantageously a self-generated consequence was. A within-subjects design was used to test the hypothesis (H1). H2 was also tested using a construal measure.
Thesis Maarten Herbrink – The Privacy Paradox: Explained using the Construal Level Theory 23 4.1.3STIMULI AND MEASUREMENTS
Respondents were confronted with three hypothetical online data disclosure situations. The first situation was regarding a fitness-app, which contained the following description “You are using a fitness-app on your smartphone. On this app you keep
track daily of what you are eating and your movement. The app asks whether you want to share your usage data. How big is the chance that you would choose to share your usage data with the app” The second situation was with regard to a fashion
webshop, with the following description “When you are visiting a fashion-webshop
you are asked whether you accept the ‘cookies’. How big is the chance that you would choose to accept these ‘cookies’?” The last situation was about a smartphone app: “You have downloaded a new app for your smartphone. The first time that you open the app, you are asked to make a choice between: 1) create a new profile. 2) Log in using your Facebook profile. How big is the chance that you would choose to log in with your Facebook profile?”
Respondents were asked to rate these situations on whether they would choose to disclose personal information or not. For situation three, respondents also had the option to choose ‘I don’t have Facebook’.
Measuring psychological distance and pro/con
After indicating their willingness to disclose, respondents were asked to generate as many possible consequences of disclosing their data for the particular situation. They were instructed to name at least one cost and one benefit of disclosing their data in the particular situation. These consequences generated by participants were used as stimuli for respondents to rate on psychological distance. Respondents were asked to rate for each consequence how they evaluated the temporal and hypothetical distance,
Thesis Maarten Herbrink – The Privacy Paradox: Explained using the Construal Level Theory 24 and whether they assessed the named consequence to be a pro or a con. Respondents did this three times over, thus, for each situation they followed the process described.
4.1.4PROCEDURE
Participants were send the link to participate in the survey. When clicking the link, respondents first got to read an introduction which contained an explanation of the context of the research, the time it would take to fill in the survey, and the fact that their data would be processed anonymously.
Then, respondents were exposed to the three situations. The sequence in which the disclosure situations were showed was randomized. Each situation consisted of the same questions and procedure. First participants were asked whether or not to would choose to disclose their personal information for the situation on a scale from 1 (definitely not) to 5 (definitely). Here after, respondents were asked to name as many consequences of disclosing their data for the current situation, and were asked to name at least one cost and one benefit. Then in the next two questions, respondents were asked to rate these self-generated costs and benefits on temporal and hypothetical distance. For temporal distance the scale to judge their self-generated consequences went from 1 (direct) to 5 (on the very long term), for hypothetical distance the scale went from 1 (very unlikely) to 5 (very likely). Lastly, participants were asked to judge their generated consequences on to how advantageously they evaluated each consequence to. This was done with a scale ranging from 1 (very disadvantageous) to 5 (very advantageous). With these questions, perceived psychological distance was measured (Boroditsky, 2000; Spence, Poortinga & Pidgeon, 2012). The construal of the self-generated consequences was measured by coding the consequences as either concrete or abstract (see 4.1.5)
Thesis Maarten Herbrink – The Privacy Paradox: Explained using the Construal Level Theory 25
Lastly, respondents were asked to fill out some demographic variables. These included gender, age and level of education. Respondents were also asked how much time they spend for personal use on the internet per day.
4.1.5DATA ANALYSIS
The data was analysed using the statistical program SPSS. The raw data was first prepared for analysis. The data was controlled for missing values and counter-indicative items. The scales items for temporal distance were recoded into different variables, such that a high score on temporal distance would also reflects a high psychological distance. Also a check for errors was conducted.
Since each respondent named multiple consequences, these consequences were used as the unit of analysis. Each consequence was reported for temporal- and hypothetical distance, and whether it was a pro or a con. New variables were created for pro/con, participant and situation.
The construal level of the consequences (costs and benefits) named by respondents was measured, to see whether benefits were construed more concrete, and the costs more abstract. In order to do this, each consequences generated by respondents was coded as either concrete or abstract, neutral and unclear costs and benefits were coded as 99 (missing value). A new variable was created: for concrete (0) or abstract (1). A consequence was rated as abstract when represented in broader categories, and linked to a superordinate purpose (Liberman & Trope, 1998; Fujita, Hernderson, Eng, Trope & Liberman, 2006). A consequence was rated as concrete when more detailed and taking place in near future (Waslak et al., 2006).
Thesis Maarten Herbrink – The Privacy Paradox: Explained using the Construal Level Theory 26
4.2RESULTS
4.2.1PSYCHOLOGICAL DISTANCE
In order to test if the benefits of disclosing personal data online were perceived as psychologically closer than the costs of disclosing this data, two mixed model analysis were conducted. The psychological distance is divided into temporal and hypothetical distance. Temporal distance is tested in the first part of H1, and hypothetical distance in the second part of H1. A mixed model analysis was performed to test the effect of costs and benefits on temporal situation and a second mixed model analysis to test these costs and benefits on hypothetical distance. A mixed model analysis seemed appropriate, because this model can include the hierarchical data structure in the analysis. It accounts for the fact that costs and benefits can not be seen as independent from each other because respondents named multiple consequences. Also, the way respondents evaluated these consequences on hypothetical and temporal distance is influenced by their personal characteristics. It was assumed in the analysis that both intercepts and slopes could vary around the model, and therefore the random intercept model was used.
In the analysis there were two independent variables (costs & benefits and situation), one dependent variable (hypothetical or temporal distance), and one contextual variable (participants). Both analysis was controlled for by gender.
Thesis Maarten Herbrink – The Privacy Paradox: Explained using the Construal Level Theory 27
Temporal distance
The analysis performed showed that the relationship between costs and benefits and temporal distance showed significant variance. Slopes did not vary across participants. The mixed model analysis showed a significant effect between costs and benefits and the temporal distance of these consequences. A significant fixed effect of costs and benefits on perceived temporal distance was found, F (1,470.777) = 16.25, p < .01. This means that whenever a
respondent indicated a consequence to be more of a benefit, the temporal – and thus psychological – distance was perceived to be smaller. In figure 4.1, it can be seen that the score for the benefits is lower on temporal distance as opposed to the costs. A lower score thus indicating
that a consequence is perceived as more likely to happen in the near future (short time span). This means that a smaller psychological distance for the benefits of disclosing personal data is perceived by respondents.
The test was controlled for gender, which showed to have no influence on the perceived level of temporal distance, F (1,58.146) = .006, p = .94. Gender thus does not influence the temporal distance perceived by respondents. The fixed effect of the situation on levels of perceived temporal distance was significant with F (1,465.380) = 23.12, p < .01. This means that the levels of temporal distance perceived vary across the three different situations. To get a better understanding of these differences, a Post-Hoc test was conducted. This showed that there was a significant positive
Figure 4.1
Effect of costs and benefits on temporal distance
Thesis Maarten Herbrink – The Privacy Paradox: Explained using the Construal Level Theory 28 difference between the ‘fitness app’ and the ‘fashion webshop’, meaning the fitness app is significantly higher (MD = .34, p < .05). Between the ‘fitness app’ and the ‘smartphone app’, the fitness app was significantly found to be higher, (MD = .6, p < .05). Lastly, between the ‘fashion webshop’ and the ‘smartphone app’, the fashion webshop was found to be significantly higher (MD = .26, p < 0.05). To give more insights into these differences between the situations, a cross categorical multilevel analysis was conducted. The results can be seen in table 4.1.
Situation Costs Benefits Total
N M (SD) N M (SD) N M (SD) 1. Fitness app 79 2.38 (1.39) 78 2.37 (1.34) 156 2.38 (1.36) 2. Fashion webshop 94 2.16 (1.2) 75 1.77 (1.12) 169 1.99 (1.18) 3. Smartphone app 88 2.16 (1.19) 88 1.36 (.85) 176 1.76 (1.11) Total 260 2.23 (1.26) 241 1.82 (1.18) 501 2.03 (1.24) Overall, the analysis showed that the perceived temporal distance of the named consequences, thus the benefit and costs, is experienced significantly smaller for the benefits of sharing personal data online than the costs. Therefore, the first part of hypothesis 1 is confirmed; the benefits of disclosing personal data online are perceived as temporally closer, whereas the costs are perceived as more temporally distant. H1 is thus supported for temporal distance.
Table 4.1
Thesis Maarten Herbrink – The Privacy Paradox: Explained using the Construal Level Theory 29
Hypothetical distance
The relationship between costs and benefits and hypothetical distance showed significant variance. Slopes did not vary across participants.
The mixed model analysis found a significant fixed effect of hypothetical distance as a dependent variable of the costs and benefits, F (1,460.068) = 9.43, p = .002 < .05. This effect can be seen in figure 4.2. This positive effect shows that the benefits score lower on the
hypothetical distance than the costs, meaning that the benefits of disclosing personal data online are perceived as psychologically closer than the costs. This means that the chances of the benefits occurring are more likely than the costs, according to the respondents. The test was controlled for
by gender, which showed to have a significant effect on hypothetical distance, F (1,73.578) = 6.74, p = .011 < p = .05. This means that gender has an influence on the perceived hypothetical distance. Men had a score of 2.02 (SD=1.03) and women M= 1.67 (SD=.77). Women thus show a smaller hypothetical distance, in this case women thus reflect that their consequences are more probable to happen.
The fixed effect of the situations on the levels of hypothetical distance was significant, F (1,459.109) = 13.51, p < .01. This means that there is a difference in hypothetical distance between the three disclosure situations. As was also done for hypothetical distance, a Post-Hoc test was conducted. This showed that there was a
Figure 4.2
Effect of costs and benefits on hypothetical distance
Thesis Maarten Herbrink – The Privacy Paradox: Explained using the Construal Level Theory 30 significant positive difference between the ‘fitness app’ and the ‘fashion webshop’, meaning the fitness app is significantly higher (MD = .37, p < .05). Between the ‘fitness app’ and the ‘smartphone app’, hypothetical distance was significantly found to be higher for the fitness app, (MD = .28, p < .05). Lastly, between the ‘fashion webshop’ and the ‘smartphone app’, there showed not to be a significant difference (MD = .09, p = .327). To give more insights into these differences between the situations a cross categorical multilevel analysis was conducted. These results can be seen in table 4.2.
Situation Costs Benefits Total
N M (SD) N M (SD) N M (SD) 1. Fitness app 79 2.13 (1.05) 78 2.08 (.84) 157 2.1 (.96) 2. Fashion webshop 94 1.78 (.87) 75 1.6 (.78) 169 1.7 (.83) 3. Smartphone app 88 1.99 (.97) 89 1.54 (.78) 177 1.76 (.91) Total 261 1.95 (.97) 242 1.73 (.83) 503 1.85 (.91) Overall, the analysis showed that the consequences of sharing personal data online addressed as benefits by respondents are rated significantly smaller on hypothetical distance than the costs of disclosing this data. Therefore, the findings of this analysis support the second part of hypothesis 2; the benefits of disclosing personal data online are perceived as hypothetically closer, whereas the costs are perceived as more hypothetically distant. H1 is thus supported for hypothetical distance. Since H1 was also supported for temporal distance, we can conclude that H1 is supported for both temporal and hypothetical distance.
Table 4.2
Thesis Maarten Herbrink – The Privacy Paradox: Explained using the Construal Level Theory 31 4.2.2CONSTRUAL LEVEL MEASURE
Just as done for psychological distance, a mixed model analysis was performed on the data. This again accounts for the fact that the consequences can not be seen separately from each other. The dependent variable was the abstractness/concreteness of the consequences, the two independent variables were the costs & benefits and situation and there was one contextual variable (participants). The analysis was controlled for by gender. It was assumed in the analysis that both intercepts and slopes could vary around the model, and therefore the random intercept model was used.
The relationship between costs and benefits and construal showed significant variance. Slopes did not vary across participants.
The mixed model analysis showed that the fixed effect of the costs and benefits on the concreteness or abstractness showed to be significant, F
(1,461.737) = 154.9, p < .01. In figure 4.3, it can be seen that the costs of the disclosing of personal data online were construed much higher on the scale from concrete – abstract. This means that the benefits are thus construed more concrete than the costs of disclosing personal data online.
The analysis was controlled for gender, which showed to have no influence on the level of construal, F (1,71.636) = .62, p = .367. Gender thus has no influence on the way people construed the costs and benefits.
The fixed effect of situation on the way costs and benefits are construed Figure 4.3
Effect of costs and benefits on construal
Thesis Maarten Herbrink – The Privacy Paradox: Explained using the Construal Level Theory 32 showed to be significant, F (1,461.935) = 17, p < .01. The way costs and benefits were construed thus varied across situations. Just like for temporal and hypothetical distance, a Post-Hoc test was conducted to get a better understanding of these differences. A significant positive difference between the ‘fitness app’ and the
‘fashion webshop’ was found, meaning the fitness app is significantly higher (MD
= .19, p < .05). Between the ‘fitness app’ and the ‘smartphone app’, the fitness app was significantly found to be higher, (MD = .18, p < .05). Lastly, between the ‘fashion webshop’ and the ‘smartphone app’, there showed not to be a significant difference (MD = .02, p = -.743). To give more insights into these differences between the situations a cross categorical multilevel analysis was conducted. These results can be seen in table 4.3. A higher score on a scale from 0-1 means that there is a higher level of abstractness perceived.
Situation Costs Benefits Total
N M (SD) N M (SD) N M (SD) 1. Fitness app 72 .82 (.39) 76 .24 (.43) 148 .52 (.5) 2. Fashion webshop 93 .48 (.5) 73 .1 (.3) 166 .31 (.47) 3. Smartphone app 86 .57 (.5) 87 .09 (.3) 173 .33 (.47) Total 251 .61 (.49) 236 .14 (.35) 487 .38 (.49) The outcomes of this analysis could be used to – for the most part – confirm H2. If we see the benefits as the actual disclosure situations, as opposed to the costs of disclosure as thinking about privacy concerns, hypothesis 2 can be confirmed based on these findings: in disclosure situations (benefits), people hold a concrete mindset, opposed to an abstract mindset when thinking about privacy concerns (costs). Thus,
Table 4.3
Thesis Maarten Herbrink – The Privacy Paradox: Explained using the Construal Level Theory 33 before looking at study 2, hypothesis 2 can be confirmed. Study 2 can serve as additional evidence for confirming H2.
4.3DISCUSSION
In this study it was expected that the benefits of disclosing personal data online would be perceived as psychologically closer than the costs of disclosing this data. The psychological distance was measured for temporal and hypothetical distance. Respondents were faced with three hypothetical disclosure situations situations in which they would or would not share their data online. Afterwards participants generated consequences of sharing this data in the disclosure situations. These self-generated consequences were then rated by respondents on temporal and hypothetical distance. Lastly they indicated how (dis)advantageous a named consequence was. The outcomes showed that participants, as expected, perceived the benefits of disclosing personal data online as psychologically closer than the costs of doing so. For both temporal and hypothetical distance, participants indicated that they perceived the benefits to be significantly psychologically closer than the costs of disclosing personal data online. This confirms H1. Another thing that was found is that for the fitness app, the psychological distance was perceived the closest – for both temporal and hypothetical distance – in relation to the other two disclosure situations. This could be explained by the fact that the fitness app is a situation that occurs less often in real life situation opposed to accepting cookies and logging in via a Facebook account. The fact that people are not very familiar with the negative consequences of this might lead them to follow their benefits.
The outcomes from the construal analysis serve as additional evidence for the existence of the Construal Level Theory. Respondents showed that they construed the
Thesis Maarten Herbrink – The Privacy Paradox: Explained using the Construal Level Theory 34 benefits of disclosing personal data online as more concrete, opposed to construing the costs of doings so as more abstract. This confirms H2, and is in line with previous research (Trope & Liberman, 2003; Liberman & Trope, 1998; Liberman, Sagristano & Trope, 2002). The psychological distance showed to be an important predictor of the construals used, which showed to be congruent with each other. This is in line with previous research (Trope & Liberman 2003; Liberman & Trope, 2006; Liberman, Sagristano & Trope, 2002).
These findings can be used to explain the privacy paradox. It seems that the dichotomy between privacy concerns and actual disclosure behaviour is caused by consumers’ irrationality. People tend to choose their psychologically closer and proximal benefits of disclosure over their psychologically more distant risks of disclosure. The benefits are construed as more concrete, whereas the costs are construed more abstract – which leads people to follow their concrete benefits.
Thesis Maarten Herbrink – The Privacy Paradox: Explained using the Construal Level Theory 35
5:STUDY2
5.1METHOD
5.1.1RESPONDENTS
In the study 76 respondents participated. These respondents were reached out to via Facebook, e-mail and Whatsapp. Out of the 76 people that started the survey, 65 completely finished the survey, which equals a completion rate of 86%. Out of these 65 respondents, 28 were male and 37 female. Looking at the age of the respondents, 81.5% (N=40) of the respondents was 25 years or younger, a remarkable big group. Education wise, 61.5% (N=40) of the respondents have a current or finished university degree. Time spent on the internet for personal use per day was for over 70% (N=48) between 1 and 4 hours per day. All demographic statistics of the respondents can be found in the appendix (F).
5.1.2DESIGN STUDY
The goal of study is to discover whether in typical disclosure situations, people hold a concrete mindset when making decisions about disclosure of personal data online (H2), opposed to a more abstract one when filling out questions about their privacy concerns. This was measured through a survey with two research groups. One group was asked several online privacy-related questions, whereas the other group was put in three hypothetical disclosure situations of personal data. The goal of putting respondents into these groups is to test the assumption that answering privacy questions would lead to a more abstract mindset, and answering questions about potential data disclosure would lead to a concrete mindset. Afterwards both groups answered questions which could determine their mindset – concrete or abstract.
Thesis Maarten Herbrink – The Privacy Paradox: Explained using the Construal Level Theory 36 5.1.3MEASUREMENTS
Disclosure situations group
For the first part of the survey respondents were randomly divided into two groups. The first group was tested on to what extent they are willing to disclose personal data, when put in hypothetical disclosure situations. There were three situations used – identical to those used in study 1. The first situation was regarding a fitness-app, it contained the following description “You are using a fitness-app on your smartphone.
On this app you keep track daily of what you are eating and your movement. The app asks whether you want to share your usage data. How big is the chance that you would choose to share your usage data with the app?” The second situation was with
regard to a fashion webshop, with the following description “When you are visiting a
fashion-webshop you are asked whether you accept the ‘cookies’. How big is the chance that you would choose to accept these ‘cookies’?” The last situation was
about a smartphone app: “You have downloaded a new app for your smartphone. The
first time that you open the app, you are asked to make a choice between: 1) create a new profile. 2) Log in using your Facebook profile. How big is the chance that you would choose to log in with your Facebook profile?” Respondents were asked
whether they would choose to disclose their personal data on a 5-point Likert scale. For situation three, respondents also had the option to choose ‘I don’t have Facebook’.
Privacy concerns group
The second group was confronted with questions with regard to their concerns about their privacy, especially their internet privacy concerns. The questions for this group were built upon different constructs. The constructs used are Willingness to act, Risk
Thesis Maarten Herbrink – The Privacy Paradox: Explained using the Construal Level Theory 37
beliefs and Confidence and enticement beliefs. These constructs and items were taken
from Dinev and Hart (2006), who showed it to be a good measure of online privacy -concerns. All constructs and sub-constructs can be seen in table 5.1. The items can be found in appendix C. The scores generated on these constructs reflected the concerns respondents held towards the specific subjects.
Construct category Construct Acronym Definition Willingness to act Willingness to provide
personal information to transact on the internet
PPIT Willingness to provide personal information required to complete transactions on the Internet.
Risk beliefs Perceived internet privacy risk
PR Perceived risk of opportunistic behaviour related to the disclosure of personal information submitted by internet users in general
Internet privacy concerns
PC Concerns about
opportunistic behaviour related to the personal information
submitted over the Internet by the respondent in particular.
Confidence and enticements beliefs
Internet trust T Trust beliefs reflecting confidence that personal information submitted to Internet websites will be handled competently, reliably, and safely. Personal internet interest PI Personal interest or cognitive attraction to Internet content overriding privacy concerns. Measuring mindset
After exposing the respondents in both research groups to these different conditions, both groups were faced with the same questions in order to measure the Table 5.1
Thesis Maarten Herbrink – The Privacy Paradox: Explained using the Construal Level Theory 38 concreteness/abstractness of their mindset. The questions used to measure mindset were deducted from Fujita et al (2006), who asked participants to generate superordinate category labels (high-level construals) or subordinate exemplars (low-level construals) for a variety of common objects. For example, the event recycling, followed by two alternate descriptions of the event. One referring to global, high-level concerns or motives in the situation (caring for the environment), whereas the other was aimed at a concrete, low-level action (bagging paper, glass and cans). Respondents were then asked to choose one of the two alternatives that described the task best. A total of 8 items was used. All items including translations to Dutch can be found in appendix D. These items generated a score that showed whether respondents evaluated the activities described as being more abstract or concrete, which showed whether respondents were holding either a more abstract or concrete mindset.
5.1.4PROCEDURE
Participants were send the link to participate in the survey. When clicking the link, respondents first got to read an introduction which contained an explanation of the context of the research, the time the research would take, and the fact that their data would be processed anonymously.
Then, participants were randomly assigned to either group one or group two. The first group (disclosure situations group) faced three disclosure situations of personal data online. Per situation participants were asked whether or not they would choose to disclose their personal information for the situation on a scale from 1 (definitely not) to 5 (definitely). These answers reflected their willingness to disclose. After having filled out these questions, participants filled out the mindset measure. The procedure for group two (privacy concerns group) lead the respondents to
Thesis Maarten Herbrink – The Privacy Paradox: Explained using the Construal Level Theory 39 five blocks of questions. First the construct willingness to act was measured, which asked: “to what extent are you willing to use the internet for the following activities”. Below, four internet usage actions were mentioned. The respondents were asked to grade them on a scale from 1 (not willing at all) to 5 (very willing).
Then the construct Risk Beliefs was measured, which consisted out of 8 items regarding the worries of respondents about potential misuse and unforeseen use of their personal information submitted online. The first four items were answered on a scale from very low risk (1) to very high risk (5), and the last four on not concerned at all (1) to very concerned (5).
Then Confidence and enticement beliefs were evaluated by respondents. This was done on a 5-point Likert scale from totally agree (1) to totally disagree (5). It contained questions about whether respondents felt internet sites were safe environments, which contained 3 items (Internet trust). The last questions were regarding the personal internet interest: does their need for information from the internet override the concerns about disclosure? This was measured with 3 items. After having filled out these questions, participants filled out the mindset measure. Lastly, respondents in both groups were asked to fill out some demographic variables. These included gender, age and level of education. Respondents were also asked how much time they spend for personal use on the internet per day.
5.1.5DATA ANALYSIS
The data was analysed using the statistical program SPSS. The raw data first had to be prepared for analysis. First of all, there was a check conducted of frequencies to examine if there were any errors in the data. These errors were dealt with by excluding cases list wise. This means that only cases that had no missing data in any
Thesis Maarten Herbrink – The Privacy Paradox: Explained using the Construal Level Theory 40 variable were analysed. Also, the data was checked for counter-indicative variables. Also, a new variable was created to account for the total the mindset respondents held (concrete or abstract) and the group respondents were in (1 or 2). New variables were also created for Willingness’ to act and for Risk Beliefs by grouping items. Also a reliability analysis was ran on the constructs of the privacy concerns.
5.2RESULTS Privacy concerns
Before testing the effect of the two groups on the mindset of respondents, the reliability of the three different constructs among privacy concerns was measured. A reliability analyses of all items of the constructs Willingness to act, Risk Beliefs and
Confidence and enticements beliefs was performed. The construct Willingness to act
had a score above .7, after deleting the first item (first α = .667, after removing α = .736). The Cronbach’s Alpha for Risk beliefs was good with α = .804. Confidence
and enticement beliefs however represented a Cronbach’s Alpha of α = .335. This
construct is thus unreliable and it was not considered in the rest of the analysis. Since the first two constructs are reliable, it is interesting to check if consumers reported high concerns. Looking at the outcomes, the risk beliefs were M = 3.7 (SD = .51), which thus showed concern of the respondents about their online privacy on a 5-point scale. The willingness to act (M = 2.99, SD = .99) showed to be quite neutral, leaning towards 3 on a 5-point scale of whether to disclose or not.
Mindset
A one way Anova was performed with the mindset score as the dependent variable
and the group respondents were in – privacy concerns or disclosure situations – as the independent variable. The differences between groups was found to be insignificant, F
